Canadian mortality clock: Technical notes

On this page

• Data sources
• Assumptions and limitations
• Calculating estimated mortality in Canada
• Visualizing linear regression calculations

Data sources

Links and explanations of data used for mortality and population.

Mortality

Using Statistics Canada's Table 13-10-0394-01, we extracted data on the top causes of death among people living in Canada between 2015 and 2024. Data included stratifications by age and sex.

Causes of death are categorized using the 10th revision of the International statistical classification of diseases and related health problems (ICD-10) ICD-10 is the international standard for classifying diagnoses. The ICD-10 uses a system of letters and numbers (which form a code) to represent different causes of death. It allows data on morbidity to be more easily stored, managed, and analyzed. More information can be found at the ICD-10 website.

Throughout this Canadian mortality clock, causes of deaths have been translated from their ICD-10 codes and official descriptions to language that is more well-known. For example, for "Malignant neoplasms [C00-C97]", we use the more well-known description of "Cancers".

Population

We used Statistics Canada's Table 17-10-0005-01 to extract the estimated number of people living in Canada on July 1 for each year between 2015 and 2025. These data are broken down by age and by sex as well.

Assumptions and limitations

In addition to those noted in the Statistics Canada data sources, this data tool makes its own assumptions and has its own limitations.

• The apportioning method (assumption of an "assembly line") was used to estimate the number of deaths that have occurred throughout the day and the year. This method assumes that deaths occur at a consistent pace. This is not accurate to reality, because deaths occur sporadically. Additionally, the frequencies of deaths by many causes fluctuate depending on the time of day, the time of year, or both.
• We made predictions using linear regression and by estimating values that do not appear in a dataset (extrapolation). This results in high levels of uncertainty in the estimated values. The estimated values included in this data tool are illustrative only, and are likely imprecise.
• It is simplistic to use only population change to estimate the number of deaths by specific causes within specific population groups in a given year. This does not reflect the many factors which contribute to the number of deaths by different causes and among different population groups year-to-year.

Calculating estimated deaths in Canada

To calculate the estimated total number of deaths in Canada, we first computed a linear regression of the number of deaths as a function of the country's population over 10 years. This linear regression was calculated for each stratification and for each of the top 10 causes of death within each stratification in 2024. We used the resulting linear functions to estimate the projected number of people who would die in 2026 based on the latest population estimates.

We used a different method for:

• linear regressions that projected a negative number of deaths for a cause within a stratification
• estimates of the number of deaths by "All other causes"

In this alternative method, we multiplied the number of deaths related to that cause by the population growth factor for the specific stratification between 2024 and 2025 (2025-estimate divided by the 2024-estimate).

Finally, to calculate the estimated amount of time elapsed between deaths by each cause, we divided the number of seconds in a years (assuming days of precisely 24 hours, and a year of 365.2422 days) by the total numbers of deaths.

Estimates using linear regression

To find the line of best fit for the number of deaths as a function of population, we first calculated the average population and the average number of deaths between 2015 and 2024 for each of the top 10 causes of death for each population breakdown.

$Average\; population\; (<mi\; style="border-top:1px\; solid\; black;">x</mi>)=\frac{\sum _{i=1}^{n}xi}{n}$

With the average death and population values, we could then find the slope of the line of best fit using the least squares method, as follows:

$Slope\; (<mi>m</mi>)=\frac{\sum _{i=1}^n\left((xi-x)\right)\left((di-d)\right)}{\sum _{i=1}^n{\left((xi-x)\right)}^2}$

Note:

• i: individual observation (one year of death or population data)
• n: number of observations (total number of years of death or population data)

Next, we calculated the intercept for each line of best fit, using the following equation:

$Intercept\; (<mi>b</mi>)=d-mx$

Finally, we could predict the number of deaths in 2026 as a function of the population in 2025, using the calculated line of best fit, according to the following formula:

$Predicted\; deaths\; in\; 2026\; (<mi>d</mi><mi\; style="font-size:12px;\; position:relative;\; bottom:-2px;">2026</mi>)=mx\mathrm{2025}+b$

Estimations based on most recent death data and population growth

As noted, any time the predicted number of deaths was negative (and to predict deaths by "All other causes"), we instead estimated the number of deaths in 2026 based on the number in 2024 and the change in population between 2024 and 2025. To do so, we used the following formula:

$Estimated\; deaths\; in\; 2026\; (<mi>d</mi><mi\; style="font-size:12px;\; position:relative;\; bottom:-2px;">2026</mi>)=d\mathrm{2024}\frac{x\mathrm{2025}}{x\mathrm{2024}}$

Calculating time between deaths by cause

After estimating the number of deaths for each cause within each stratification, we calculated the average amount of time between deaths for each estimate. We did this by dividing the number of seconds in a years (according to a year of 365.2422 days) by the total numbers of deaths:

$Time\; between\; deaths=\frac{60\times 60\times 24\times 365.2422}{d\mathrm{2026}}$

Visualizing linear regression calculations

Figure 1: Linear regressions (where applicable) for the top 10 causes of death among

Figure 1: Text description